Pascual and López-Barneo reply:
We have reported that GDNF is required for adult catecholaminergic neuron survival1. We generated heterozygous mice with a loxP-flanked (floxed) Gdnf allele and estrogen receptor–associated Cre recombinase (Esr1-Cre, F/− mice) in which the Gdnf gene was ablated in adulthood (2 months of age) by treatment with tamoxifen (TMX). The mice showed a progressive hypokinesia and selective decrease of brain tyrosine hydroxylase (Th) mRNA 7 months after GDNF downregulation (to ∼40% of the normal striatal protein content). This was accompanied by pronounced catecholaminergic cell death affecting the locus coeruleus, the substantia nigra and the ventral tegmental area. Kopra et al.2 have challenged our results by using a triple approach to inhibit striatal GDNF production. Several issues need to be considered here. First, the Nestin-Cre–driven deletion of the Gdnf floxed allele is an interesting model because, as the animals can reach adulthood, it demonstrates dopaminergic nigrostriatal neuron survival in the complete absence of GDNF. However, in this mouse line, Gdnf deletion occurs during development and the possibility of embryonic compensation therefore cannot be dismissed. Indeed, although it is well established that GDNF is transiently required during development for defining adult subtypes of Th+ midbrain neurons3, animals without GDNF are born with a normal number of dopaminergic neurons, which makes the hypothesis of embryonic compensation highly plausible. Second, Kopra et al.2 also used an Esr1-Cre–driven recombination model to delete the floxed Gdnf alleles in adult animals. However, their study is hampered by the fact that, for unknown reasons, Gdnf deletion was not induced by TMX treatment. In contrast, they seemed to observe constitutive striatal Cre activation that, nonetheless, did not induce a substantial reduction of striatal GDNF protein. The behavior exhibited by this ESR1-Cre mouse line is unusual, as we have shown with several mouse models carrying floxed alleles that substantial recombination only occurs after TMX treatment1,4,5. The efficiency of TMX-activated Esr1-Cre is highly variable, and the distinct effects observed with this Cre line could be a result of variations in the genetic background (F1 hybrid C57bl6/129SvJ in Pascual et al.1 and mixed 129Ola/ICR/C57bl6 in Kopra et al.2). Third, the final model employed by Kopra et al.2 was the intrastriatal injection of AAV5-Cre in F/F mice (with two floxed Gdnf alleles. The reported efficiency of Gdnf deletion (measured as Gdnf mRNA in the entire striatum 30 d after AAV5 treatment) was highly variable (between 20 and 80% of the control value) and the amount of striatal GDNF protein was maintained above 50% of the normal level. Hence, one could argue that GDNF content was not sufficiently reduced to elicit neurodegeneration in these mice with a mixed genetic background, and that they might be more resistant to GDNF deficiency than other strains. In addition, the possibility that tissue damage produced by stereotaxic AAV injections could have activated glial cells to produce trophic factors that compensate for the lack of GDNF cannot be ruled out6.
This is a preview of subscription content, access via your institution
